A Fishervoice-SVM language identification system

Authors:
Paula Lopez-Otero;Laura Docio-Fernandez;Carmen Garcia-Mateo
Affiliations:
Multimedia Technologies Group, E.E. Telecomunicación, Universidade de Vigo, Vigo, Spain;Multimedia Technologies Group, E.E. Telecomunicación, Universidade de Vigo, Vigo, Spain;Multimedia Technologies Group, E.E. Telecomunicación, Universidade de Vigo, Vigo, Spain
Venue:
PROPOR'12 Proceedings of the 10th international conference on Computational Processing of the Portuguese Language
Year:
2012

Citing 3
Cited 0

A Tutorial on Support Vector Machines for Pattern Recognition

Data Mining and Knowledge Discovery
Rapid and brief communication: Face recognition based on 2D Fisherface approach

Pattern Recognition
LIBSVM: A library for support vector machines

ACM Transactions on Intelligent Systems and Technology (TIST)

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper, a language identification system is described that implements the Fishervoice approach in order to reduce the dimensionality of the data. Fishervoice performs two-dimensional Principal Component Analysis (2D-PCA) and Linear Discriminant Analysis (LDA) to project the data into a discriminative subspace. After this transformation the speech utterances are transformed into supervectors and classified by means of a Support Vector Machine (SVM). Experiments performed on KALAKA-2 database, which includes speech in Spanish, Catalan, English, Basque, Galician and Portuguese, show that the Fishervoice-SVM system achieves good identification results while reducing dramatically the number of features needed to represent the speech utterances.